Cancel Tapis Job

Cancel Tapis Job#

how to stop jobs using Tapis, specifically with the Python client tapipy

Once you’ve submitted a job using Tapis, you may need to stop or cancel it—for example, if you detect a mistake in the input, want to halt a long-running job, or need to resubmit with different parameters. Tapis provides the ability to cancel a job that is queued or currently running.

NOTE: This page is not an interactive Jupyter Notebook.

Stopping a Job via CLI#

If you’re using the Tapis CLI, you can cancel a job using the following command:

tapis jobs cancel $JOB_UUID

Replace $JOB_UUID with the actual UUID of your submitted job.

Stopping a Job Using tapipy#

To stop (cancel) a job programmatically using Python and the tapipy client, use the cancelJob() method on the jobs service. Here’s a simple example:

# Assuming *t* is your authenticated Tapis client
job_uuid = 'your-job-uuid-here'

# Cancel the job
t.jobs.cancelJob(jobId=job_uuid)
print(f"Job {job_uuid} has been cancelled.")

Notes:#

  • The job must be in a PENDING, PROCESSING_INPUTS, or RUNNING state for the cancel operation to succeed.

  • Canceling a job will attempt to terminate the corresponding job on the execution system (e.g., Stampede3 or Frontera), and it will no longer produce output.

Monitoring Cancellations#

After calling cancelJob, you can monitor the job’s status using:

job_status = t.jobs.getJob(jobId=job_uuid).status
print(f"Current job status: {job_status}")

Once canceled, the job’s final status will typically show as CANCELLED.

When to Use This#

Use job cancellation when:

  • You submitted incorrect inputs.

  • The job is taking too long due to misconfiguration.

  • You want to free up resources for another run.

  • You are debugging or testing job workflows.

Tip: When Jobs Can’t Be Cancelled#

Some jobs may already be in a terminal state (FINISHED, FAILED, CANCELLED) and cannot be stopped. You can pre-check status like this:

status = t.jobs.getJob(jobId=job_id).status
if status in ['FINISHED', 'FAILED', 'CANCELLED']:
    print(f"Job is already in a terminal state: {status}")
else:
    cancel_tapis_job(t, job_id)
Contents